[PATCH 3/4] hw/block/nvme: supply dw1 for aen result

2022-10-20 Thread clay.mayers 
From: Clay Mayers 

cqe.dw1 AEN is sometimes required to convey the NSID of the log page
to read.  This is the case for the zone descriptor changed log
page.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 19 +++
 hw/nvme/nvme.h   |  2 ++
 hw/nvme/trace-events |  2 +-
 include/block/nvme.h |  4 +++-
 4 files changed, 17 insertions(+), 10 deletions(-)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 3ffd0fb469..c7ee54ef5e 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1445,6 +1445,7 @@ static void nvme_process_aers(void *opaque)
 result->event_type = event->result.event_type;
 result->event_info = event->result.event_info;
 result->log_page = event->result.log_page;
+req->cqe.dw1 = cpu_to_le32(event->result.nsid);
 g_free(event);
 
 trace_pci_nvme_aer_post_cqe(result->event_type, result->event_info,
@@ -1455,11 +1456,12 @@ static void nvme_process_aers(void *opaque)
 }
 
 static void nvme_enqueue_event(NvmeCtrl *n, uint8_t event_type,
-   uint8_t event_info, uint8_t log_page)
+   uint8_t event_info, uint8_t log_page,
+   uint32_t nsid)
 {
 NvmeAsyncEvent *event;
 
-trace_pci_nvme_enqueue_event(event_type, event_info, log_page);
+trace_pci_nvme_enqueue_event(event_type, event_info, log_page, nsid);
 
 if (n->aer_queued == n->params.aer_max_queued) {
 trace_pci_nvme_enqueue_event_noqueue(n->aer_queued);
@@ -1471,6 +1473,7 @@ static void nvme_enqueue_event(NvmeCtrl *n, uint8_t 
event_type,
 .event_type = event_type,
 .event_info = event_info,
 .log_page   = log_page,
+.nsid   = nsid,
 };
 
 QTAILQ_INSERT_TAIL(&n->aer_queue, event, entry);
@@ -1505,7 +1508,7 @@ static void nvme_smart_event(NvmeCtrl *n, uint8_t event)
 return;
 }
 
-nvme_enqueue_event(n, NVME_AER_TYPE_SMART, aer_info, NVME_LOG_SMART_INFO);
+nvme_enqueue_event(n, NVME_AER_TYPE_SMART, aer_info, NVME_LOG_SMART_INFO, 
0);
 }
 
 static void nvme_clear_events(NvmeCtrl *n, uint8_t event_type)
@@ -5823,7 +5826,7 @@ static uint16_t nvme_ns_attachment(NvmeCtrl *n, 
NvmeRequest *req)
 if (!test_and_set_bit(nsid, ctrl->changed_nsids)) {
 nvme_enqueue_event(ctrl, NVME_AER_TYPE_NOTICE,
NVME_AER_INFO_NOTICE_NS_ATTR_CHANGED,
-   NVME_LOG_CHANGED_NSLIST);
+   NVME_LOG_CHANGED_NSLIST, 0);
 }
 }
 
@@ -6964,7 +6967,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_REGISTER,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -6981,7 +6984,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_VALUE,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -7026,7 +7029,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_REGISTER,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -7043,7 +7046,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_VALUE,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
diff --git a/hw/nvme/nvme.h b/hw/nvme/nvme.h
index ae65226150..2b7997e4a7 100644
--- a/hw/nvme/nvme.h
+++ b/hw/nvme/nvme.h
@@ -477,6 +477,8 @@ typedef struct NvmeCtrl {
 uint64_tdbbuf_eis;
 booldbbuf_enabled;
 
+boolzdc_event_queued;
+
 struct {
 MemoryRegion mem;
 uint8_t  *buf;
diff --git a/hw/nvme/trace-events b/hw/nvme/trace-events
index 337927e607..86c01f8762 100644
--- a/hw/nvme/trace-events
+++ b/hw/nvme/trace-events
@@ -80,7 +80,7 @@ pci_nvme_aer_masked(uint8_t type, uint8_t mask) "type 
0x%"PRIx8" mask 0x%"PRIx8"
 pci_nvme_aer_post_cqe(uint8_t typ, uint8_t info, uint8_t log_page) "type 
0x%"PRIx8" info 0x%"PRIx8" lid 0x%"PRIx8""
 pci_nvme_ns_attachment(uint16_t cid, uint8_t sel) "cid %"PRIu16", 
sel=0x%

[PATCH 4/4] hw/block/nvme: add zone descriptor changed AEN

2022-10-20 Thread clay.mayers 
From: Clay Mayers 

If a namespace's param.zoned.finish_time is non-zero,
controllers register with the namespace to be notified
when entries are added to its zone-descriptor-changed
log page.  If the zone-descriptor-changed aen is enabled,
this will cause an AEN to be sent from that controller.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 62 +++-
 hw/nvme/ns.c |  1 +
 hw/nvme/nvme.h   |  9 +++
 include/block/nvme.h |  2 ++
 4 files changed, 73 insertions(+), 1 deletion(-)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index c7ee54ef5e..f1cfa272b4 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1519,6 +1519,52 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_zdc_watch(NvmeCtrl *n, NvmeNamespace *ns, NvmeNotifyFnc fnc)
+{
+NvmeZdcNotify *watcher = g_malloc0(sizeof(*watcher));
+
+watcher->n = n;
+watcher->notify = fnc;
+QTAILQ_INSERT_TAIL(&ns->zdc_watchers, watcher, entry);
+}
+
+static void nvme_zdc_unwatch(NvmeCtrl *n, NvmeNamespace *ns)
+{
+NvmeZdcNotify *watcher;
+
+QTAILQ_FOREACH(watcher, &ns->zdc_watchers, entry) {
+if (watcher->n == n) {
+QTAILQ_REMOVE(&ns->zdc_watchers, watcher, entry);
+break;
+}
+}
+}
+
+static void nvme_zdc_notify(NvmeNamespace *ns)
+{
+NvmeZdcNotify *watcher;
+
+QTAILQ_FOREACH(watcher, &ns->zdc_watchers, entry) {
+(*watcher->notify)(watcher->n, ns);
+}
+}
+
+static void nvme_zdc_aen(NvmeCtrl *n, NvmeNamespace *ns)
+{
+g_assert(n->id_ctrl.oaes & (1 << 27));
+
+if (!NVME_AEC_ZONE_CHANGED(n->features.async_config)) {
+return;
+}
+
+if (!n->zdc_event_queued) {
+n->zdc_event_queued = true;
+nvme_enqueue_event(n, NVME_AER_TYPE_NOTICE,
+NVME_AER_INFO_NOTICE_ZONE_DESC_CHANGED,
+NVME_LOG_CHANGED_ZONE, ns->params.nsid);
+}
+}
+
 static void nvme_zdc_list(NvmeNamespace *ns, NvmeZoneIdList *zlist, bool reset)
 {
 NvmeZdc *zdc;
@@ -1554,6 +1600,7 @@ static void nvme_check_finish(NvmeNamespace *ns, 
NvmeZoneListHead *list)
 zdc->zone = zone;
 zone->zdc_entry = zdc;
 QTAILQ_INSERT_TAIL(&ns->zdc_list, zdc, entry);
+nvme_zdc_notify(ns);
 }
 } else if (zone->finish_ms != INT64_MAX) {
 timer_mod_anticipate(ns->active_timer, zone->finish_ms);
@@ -4722,6 +4769,14 @@ static uint16_t nvme_changed_zones(NvmeCtrl *n, uint8_t 
rae, uint32_t buf_len,
 return NVME_INVALID_NSID | NVME_DNR;
 }
 nvme_zdc_list(ns, &zlist, !rae);
+if (!rae) {
+n->zdc_event_queued = false;
+nvme_clear_events(n, NVME_AER_TYPE_NOTICE);
+/* send a new aen if there are still zdc entries */
+if (!QTAILQ_EMPTY(&ns->zdc_list)) {
+nvme_zdc_notify(ns);
+}
+}
 
 return nvme_c2h(n, ((uint8_t *)&zlist) + off, trans_len, req);
 }
@@ -5808,6 +5863,7 @@ static uint16_t nvme_ns_attachment(NvmeCtrl *n, 
NvmeRequest *req)
 return NVME_NS_NOT_ATTACHED | NVME_DNR;
 }
 
+nvme_zdc_unwatch(n, ns);
 ctrl->namespaces[nsid] = NULL;
 ns->attached--;
 
@@ -7535,7 +7591,7 @@ static void nvme_init_ctrl(NvmeCtrl *n, PCIDevice 
*pci_dev)
 
 id->cntlid = cpu_to_le16(n->cntlid);
 
-id->oaes = cpu_to_le32(NVME_OAES_NS_ATTR);
+id->oaes = cpu_to_le32(NVME_OAES_NS_ATTR | NVME_OAES_ZDC);
 id->ctratt |= cpu_to_le32(NVME_CTRATT_ELBAS);
 
 id->rab = 6;
@@ -7652,6 +7708,10 @@ void nvme_attach_ns(NvmeCtrl *n, NvmeNamespace *ns)
 
 n->dmrsl = MIN_NON_ZERO(n->dmrsl,
 BDRV_REQUEST_MAX_BYTES / nvme_l2b(ns, 1));
+
+if (ns->params.fto) {
+nvme_zdc_watch(n, ns, nvme_zdc_aen);
+}
 }
 
 static void nvme_realize(PCIDevice *pci_dev, Error **errp)
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index 25cd490c99..5629b61302 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -241,6 +241,7 @@ static void nvme_ns_zoned_init_state(NvmeNamespace *ns)
 QTAILQ_INIT(&ns->closed_zones);
 QTAILQ_INIT(&ns->full_zones);
 QTAILQ_INIT(&ns->zdc_list);
+QTAILQ_INIT(&ns->zdc_watchers);
 
 zone = ns->zone_array;
 for (i = 0; i < ns->num_zones; i++, zone++) {
diff --git a/hw/nvme/nvme.h b/hw/nvme/nvme.h
index 2b7997e4a7..5499105e7b 100644
--- a/hw/nvme/nvme.h
+++ b/hw/nvme/nvme.h
@@ -91,6 +91,14 @@ static inline NvmeNamespace *nvme_subsys_ns(NvmeSubsystem 
*subsys,
 #define NVME_NS(obj) \
 OBJECT_CHECK(NvmeNamespace, (obj), TYPE_NVME_NS)
 
+typedef void (*NvmeNotifyFnc)(NvmeCtrl *n, NvmeNamespace *ns);
+
+typedef struct NvmeZdcNotify {
+QTAILQ_ENTRY(NvmeZdcNotify) entry;
+NvmeNotifyFnc notify;
+NvmeCtrl *n;
+} NvmeZdcNotify;
+
 typedef struct NvmeZdc {
 QTAILQ_ENTRY(NvmeZdc) entry;
 NvmeZone *zone;
@@ -179,6 +187,7 @@ typedef struct NvmeNamespace {
 
 int

[PATCH 1/4] hw/block/nvme: add ZONE_FINISH_RECOMMENDED functionality

2022-10-20 Thread clay.mayers 
From: Clay Mayers 

Adds ns.param.zoned.finish_time, which sets the number of
seconds a zone can remain active before the zone attribute
ZONE_FINISH_RECOMMENDED is set.

This requires scanning the exp open, imp open and closed lists
of zones whenever a zone is marked as requiring finishing.  The
expectation is these lists will be short (10s of items) allowing a
simpler implementation than keeping the lists sorted.  It also
keeps the overhead during the exception of a timeout instead of
when zones change state between open and closed. For use cases
where this isn't true, finish_time should be 0 to disable this
feature (the default).

Signed-off-by: Clay Mayers 
---
 docs/system/devices/nvme.rst |  5 +
 hw/nvme/ctrl.c   | 35 +++
 hw/nvme/ns.c |  8 
 hw/nvme/nvme.h   | 18 ++
 4 files changed, 62 insertions(+), 4 deletions(-)

diff --git a/docs/system/devices/nvme.rst b/docs/system/devices/nvme.rst
index 30f841ef62..1cb0ef844c 100644
--- a/docs/system/devices/nvme.rst
+++ b/docs/system/devices/nvme.rst
@@ -212,6 +212,11 @@ The namespace may be configured with additional parameters
   the minimum memory page size (CAP.MPSMIN). The default value (``0``)
   has this property inherit the ``mdts`` value.
 
+``zoned.finish_time=UINT32`` (default: ``0``)
+  Set the time in seconds for how long a zone can be active before setting the
+  zone attribute ``Zone Finish Recommended``.  The default value (``0``)
+  disables this feature.
+
 Metadata
 
 
diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 87aeba0564..d7e9fae0b0 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1516,6 +1516,38 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_check_finish(NvmeNamespace *ns, NvmeZoneListHead *list)
+{
+int64_t now = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL);
+NvmeZone *zone;
+
+QTAILQ_FOREACH(zone, list, entry) {
+if (zone->finish_ms <= now) {
+zone->finish_ms = INT64_MAX;
+zone->d.za |= NVME_ZA_FINISH_RECOMMENDED;
+} else if (zone->finish_ms != INT64_MAX) {
+timer_mod_anticipate(ns->active_timer, zone->finish_ms);
+}
+}
+}
+
+void nvme_finish_needed(void *opaque)
+{
+NvmeNamespace *ns = opaque;
+
+nvme_check_finish(ns, &ns->exp_open_zones);
+nvme_check_finish(ns, &ns->imp_open_zones);
+nvme_check_finish(ns, &ns->closed_zones);
+}
+
+void nvme_set_active_timeout(NvmeNamespace *ns, NvmeZone *zone)
+{
+if (ns->fto_ms) {
+zone->finish_ms = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + ns->fto_ms;
+timer_mod_anticipate(ns->active_timer, zone->finish_ms);
+}
+}
+
 static inline uint16_t nvme_check_mdts(NvmeCtrl *n, size_t len)
 {
 uint8_t mdts = n->params.mdts;
@@ -1791,6 +1823,7 @@ static uint16_t nvme_zrm_finish(NvmeNamespace *ns, 
NvmeZone *zone)
 
 /* fallthrough */
 case NVME_ZONE_STATE_EMPTY:
+zone->d.za &= ~NVME_ZA_FINISH_RECOMMENDED;
 nvme_assign_zone_state(ns, zone, NVME_ZONE_STATE_FULL);
 return NVME_SUCCESS;
 
@@ -1891,6 +1924,7 @@ static uint16_t nvme_zrm_open_flags(NvmeCtrl *n, 
NvmeNamespace *ns,
 
 if (act) {
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 }
 
 nvme_aor_inc_open(ns);
@@ -3619,6 +3653,7 @@ static uint16_t nvme_set_zd_ext(NvmeNamespace *ns, 
NvmeZone *zone)
 return status;
 }
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 zone->d.za |= NVME_ZA_ZD_EXT_VALID;
 nvme_assign_zone_state(ns, zone, NVME_ZONE_STATE_CLOSED);
 return NVME_SUCCESS;
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index 62a1f97be0..b577f2d8e0 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -322,6 +322,11 @@ static void nvme_ns_init_zoned(NvmeNamespace *ns)
 ns->id_ns.nsfeat &= ~0x4;
 }
 
+ns->fto_ms = ns->params.fto * INT64_C(1000);
+if (ns->fto_ms) {
+ns->active_timer = timer_new_ms(QEMU_CLOCK_VIRTUAL, nvme_finish_needed,
+ns);
+}
 ns->id_ns_zoned = id_ns_z;
 }
 
@@ -338,6 +343,7 @@ static void nvme_clear_zone(NvmeNamespace *ns, NvmeZone 
*zone)
 nvme_set_zone_state(zone, NVME_ZONE_STATE_CLOSED);
 }
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 QTAILQ_INSERT_HEAD(&ns->closed_zones, zone, entry);
 } else {
 trace_pci_nvme_clear_ns_reset(state, zone->d.zslba);
@@ -521,6 +527,7 @@ void nvme_ns_shutdown(NvmeNamespace *ns)
 void nvme_ns_cleanup(NvmeNamespace *ns)
 {
 if (ns->params.zoned) {
+timer_free(ns->active_timer);
 g_free(ns->id_ns_zoned);
 g_free(ns->zone_array);
 g_free(ns->zd_extensions);
@@ -644,6 +651,7 @@ static Property nvme_ns_props[] = {
 DEFINE_PROP_SIZE("zoned.zrwafg", NvmeNamespace, params.zrwafg, -1),

[PATCH 2/4] hw/block/nvme: add zone descriptor changed log page

2022-10-20 Thread clay.mayers 
From: Clay Mayers 

Zones marked with ZONE_FINISH_RECOMMENDED are added to the zone
descriptor changed log page.  Once read with RAE cleared, they are
removed from the list.

Zones stay in the list regardless of what other states the zones may
go through so applications must be aware of ABA issues where finish
may be recommended, the zone freed and re-opened and now the attribute
is now clear.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 50 
 hw/nvme/ns.c |  6 ++
 hw/nvme/nvme.h   |  8 +++
 hw/nvme/trace-events |  1 +
 include/block/nvme.h |  8 +++
 5 files changed, 73 insertions(+)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index d7e9fae0b0..3ffd0fb469 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1516,15 +1516,42 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_zdc_list(NvmeNamespace *ns, NvmeZoneIdList *zlist, bool reset)
+{
+NvmeZdc *zdc;
+NvmeZdc *next;
+int index = 0;
+
+QTAILQ_FOREACH_SAFE(zdc, &ns->zdc_list, entry, next) {
+if (index >= ARRAY_SIZE(zlist->zids)) {
+break;
+}
+zlist->zids[index++] = zdc->zone->d.zslba;
+if (reset) {
+QTAILQ_REMOVE(&ns->zdc_list, zdc, entry);
+zdc->zone->zdc_entry = NULL;
+g_free(zdc);
+}
+}
+zlist->nzid = cpu_to_le16(index);
+}
+
 static void nvme_check_finish(NvmeNamespace *ns, NvmeZoneListHead *list)
 {
 int64_t now = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL);
 NvmeZone *zone;
+NvmeZdc  *zdc;
 
 QTAILQ_FOREACH(zone, list, entry) {
 if (zone->finish_ms <= now) {
 zone->finish_ms = INT64_MAX;
 zone->d.za |= NVME_ZA_FINISH_RECOMMENDED;
+if (!zone->zdc_entry) {
+zdc = g_malloc0(sizeof(*zdc));
+zdc->zone = zone;
+zone->zdc_entry = zdc;
+QTAILQ_INSERT_TAIL(&ns->zdc_list, zdc, entry);
+}
 } else if (zone->finish_ms != INT64_MAX) {
 timer_mod_anticipate(ns->active_timer, zone->finish_ms);
 }
@@ -4675,6 +4702,27 @@ static uint16_t nvme_cmd_effects(NvmeCtrl *n, uint8_t 
csi, uint32_t buf_len,
 return nvme_c2h(n, ((uint8_t *)&log) + off, trans_len, req);
 }
 
+static uint16_t nvme_changed_zones(NvmeCtrl *n, uint8_t rae, uint32_t buf_len,
+uint64_t off, NvmeRequest *req)
+{
+NvmeNamespace *ns;
+NvmeCmd *cmd = &req->cmd;
+uint32_t nsid = le32_to_cpu(cmd->nsid);
+NvmeZoneIdList zlist = { };
+uint32_t trans_len = MIN(sizeof(zlist) - off, buf_len);
+
+nsid = le32_to_cpu(cmd->nsid);
+trace_pci_nvme_changed_zones(nsid);
+
+ns = nvme_ns(n, nsid);
+if (!ns) {
+return NVME_INVALID_NSID | NVME_DNR;
+}
+nvme_zdc_list(ns, &zlist, !rae);
+
+return nvme_c2h(n, ((uint8_t *)&zlist) + off, trans_len, req);
+}
+
 static uint16_t nvme_get_log(NvmeCtrl *n, NvmeRequest *req)
 {
 NvmeCmd *cmd = &req->cmd;
@@ -4722,6 +4770,8 @@ static uint16_t nvme_get_log(NvmeCtrl *n, NvmeRequest 
*req)
 return nvme_changed_nslist(n, rae, len, off, req);
 case NVME_LOG_CMD_EFFECTS:
 return nvme_cmd_effects(n, csi, len, off, req);
+case NVME_LOG_CHANGED_ZONE:
+return nvme_changed_zones(n, rae, len, off, req);
 default:
 trace_pci_nvme_err_invalid_log_page(nvme_cid(req), lid);
 return NVME_INVALID_FIELD | NVME_DNR;
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index b577f2d8e0..25cd490c99 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -240,6 +240,7 @@ static void nvme_ns_zoned_init_state(NvmeNamespace *ns)
 QTAILQ_INIT(&ns->imp_open_zones);
 QTAILQ_INIT(&ns->closed_zones);
 QTAILQ_INIT(&ns->full_zones);
+QTAILQ_INIT(&ns->zdc_list);
 
 zone = ns->zone_array;
 for (i = 0; i < ns->num_zones; i++, zone++) {
@@ -526,8 +527,13 @@ void nvme_ns_shutdown(NvmeNamespace *ns)
 
 void nvme_ns_cleanup(NvmeNamespace *ns)
 {
+NvmeZdc *zdc;
+
 if (ns->params.zoned) {
 timer_free(ns->active_timer);
+while ((zdc = QTAILQ_FIRST(&ns->zdc_list))) {
+g_free(zdc);
+}
 g_free(ns->id_ns_zoned);
 g_free(ns->zone_array);
 g_free(ns->zd_extensions);
diff --git a/hw/nvme/nvme.h b/hw/nvme/nvme.h
index 9a54dcdb32..ae65226150 100644
--- a/hw/nvme/nvme.h
+++ b/hw/nvme/nvme.h
@@ -32,6 +32,7 @@ QEMU_BUILD_BUG_ON(NVME_MAX_NAMESPACES > NVME_NSID_BROADCAST - 
1);
 
 typedef struct NvmeCtrl NvmeCtrl;
 typedef struct NvmeNamespace NvmeNamespace;
+typedef struct NvmeZone NvmeZone;
 
 #define TYPE_NVME_BUS "nvme-bus"
 OBJECT_DECLARE_SIMPLE_TYPE(NvmeBus, NVME_BUS)
@@ -90,10 +91,16 @@ static inline NvmeNamespace *nvme_subsys_ns(NvmeSubsystem 
*subsys,
 #define NVME_NS(obj) \
 OBJECT_CHECK(NvmeNamespace, (obj), TYPE_NVME_NS)
 
+typedef struct NvmeZdc {
+QTAILQ_ENTRY(NvmeZdc) entry;
+NvmeZone *zone;
+} NvmeZdc;
+

[PATCH 0/4] hw/block/nvme: Implement ZNS finish-zone ZDC AEN

2022-10-20 Thread clay.mayers 
From: Clay Mayers 

ZNS controllers have the option to limit the time a zone can remain in
the active state.  It begins with a background process in the controller
setting the finish-zone-recommended FZR attribute for a zone.  As part of
setting this attribute, the zone's id is added to the namespace's
zone-descriptor-changed (ZDC) log page. If enabled, items added to the
ZDC log page generate a ZDC "asynchronous event notification" AEN. Optionally,
the control can induce a "zone excursion" forcing the zone into the finished
state that also generates a ZDC event.

Zone enabled applications need to properly handle ZDC events. In a real device,
the timeout is many hours making testing an application difficult.
Implemented is the generation of FZR ZDC events to speed up O/S and application
testing.

Added to the zoned NVMe command set is an optional, per-namespace timer
(zoned.finish_time) to set the FZR attr for long-lived active zones; A per
namespace ZDC log page; AEN results to including CQE.DW1 (the NSID of the ZDC
AEN) and generating a ZDC AEN if it's been enabled. Zone excursions are not
modeled.

See section 5.5 of the NVMe Zoned Namespace Command Set Specification v1.1
for more details.

Clay Mayers (4):
  hw/block/nvme: add ZONE_FINISH_RECOMMENDED functionality
  hw/block/nvme: add zone descriptor changed log page
  hw/block/nvme: supply dw1 for aen result
  hw/block/nvme: add zone descriptor changed AEN

 docs/system/devices/nvme.rst |   5 ++
 hw/nvme/ctrl.c   | 166 +--
 hw/nvme/ns.c |  15 
 hw/nvme/nvme.h   |  37 +++-
 hw/nvme/trace-events |   3 +-
 include/block/nvme.h |  14 ++-
 6 files changed, 225 insertions(+), 15 deletions(-)

-- 
2.27.0

[PATCH V2 2/4] hw/block/nvme: add zone descriptor changed log page

2022-10-21 Thread clay.mayers 
From: Clay Mayers 

Zones marked with ZONE_FINISH_RECOMMENDED are added to the zone
descriptor changed log page.  Once read with RAE cleared, they are
removed from the list.

Zones stay in the list regardless of what other states the zones may
go through so applications must be aware of ABA issues where finish
may be recommended, the zone freed and re-opened and now the attribute
is now clear.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 57 
 hw/nvme/ns.c |  6 +
 hw/nvme/nvme.h   |  8 +++
 hw/nvme/trace-events |  1 +
 include/block/nvme.h |  8 +++
 5 files changed, 80 insertions(+)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index d7e9fae0b0..5b5ceb90c2 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1516,15 +1516,42 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_zdc_list(NvmeNamespace *ns, NvmeZoneIdList *zlist, bool reset)
+{
+NvmeZdc *zdc;
+NvmeZdc *next;
+int index = 0;
+
+QTAILQ_FOREACH_SAFE(zdc, &ns->zdc_list, entry, next) {
+if (index >= ARRAY_SIZE(zlist->zids)) {
+break;
+}
+zlist->zids[index++] = zdc->zone->d.zslba;
+if (reset) {
+QTAILQ_REMOVE(&ns->zdc_list, zdc, entry);
+zdc->zone->zdc_entry = NULL;
+g_free(zdc);
+}
+}
+zlist->nzid = cpu_to_le16(index);
+}
+
 static void nvme_check_finish(NvmeNamespace *ns, NvmeZoneListHead *list)
 {
 int64_t now = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL);
 NvmeZone *zone;
+NvmeZdc  *zdc;
 
 QTAILQ_FOREACH(zone, list, entry) {
 if (zone->finish_ms <= now) {
 zone->finish_ms = INT64_MAX;
 zone->d.za |= NVME_ZA_FINISH_RECOMMENDED;
+if (!zone->zdc_entry) {
+zdc = g_malloc0(sizeof(*zdc));
+zdc->zone = zone;
+zone->zdc_entry = zdc;
+QTAILQ_INSERT_TAIL(&ns->zdc_list, zdc, entry);
+}
 } else if (zone->finish_ms != INT64_MAX) {
 timer_mod_anticipate(ns->active_timer, zone->finish_ms);
 }
@@ -4675,6 +4702,34 @@ static uint16_t nvme_cmd_effects(NvmeCtrl *n, uint8_t 
csi, uint32_t buf_len,
 return nvme_c2h(n, ((uint8_t *)&log) + off, trans_len, req);
 }
 
+static uint16_t nvme_changed_zones(NvmeCtrl *n, uint8_t rae, uint32_t buf_len,
+uint64_t off, NvmeRequest *req)
+{
+NvmeNamespace *ns;
+NvmeCmd *cmd = &req->cmd;
+uint32_t nsid = le32_to_cpu(cmd->nsid);
+NvmeZoneIdList zlist = { };
+uint32_t trans_len;
+
+if (off >= sizeof(zlist)) {
+trace_pci_nvme_err_invalid_log_page_offset(off, sizeof(zlist));
+return NVME_INVALID_FIELD | NVME_DNR;
+}
+
+nsid = le32_to_cpu(cmd->nsid);
+trace_pci_nvme_changed_zones(nsid);
+
+ns = nvme_ns(n, nsid);
+if (!ns) {
+return NVME_INVALID_NSID | NVME_DNR;
+}
+nvme_zdc_list(ns, &zlist, !rae);
+
+trans_len = MIN(sizeof(zlist) - off, buf_len);
+
+return nvme_c2h(n, ((uint8_t *)&zlist) + off, trans_len, req);
+}
+
 static uint16_t nvme_get_log(NvmeCtrl *n, NvmeRequest *req)
 {
 NvmeCmd *cmd = &req->cmd;
@@ -4722,6 +4777,8 @@ static uint16_t nvme_get_log(NvmeCtrl *n, NvmeRequest 
*req)
 return nvme_changed_nslist(n, rae, len, off, req);
 case NVME_LOG_CMD_EFFECTS:
 return nvme_cmd_effects(n, csi, len, off, req);
+case NVME_LOG_CHANGED_ZONE:
+return nvme_changed_zones(n, rae, len, off, req);
 default:
 trace_pci_nvme_err_invalid_log_page(nvme_cid(req), lid);
 return NVME_INVALID_FIELD | NVME_DNR;
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index b577f2d8e0..25cd490c99 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -240,6 +240,7 @@ static void nvme_ns_zoned_init_state(NvmeNamespace *ns)
 QTAILQ_INIT(&ns->imp_open_zones);
 QTAILQ_INIT(&ns->closed_zones);
 QTAILQ_INIT(&ns->full_zones);
+QTAILQ_INIT(&ns->zdc_list);
 
 zone = ns->zone_array;
 for (i = 0; i < ns->num_zones; i++, zone++) {
@@ -526,8 +527,13 @@ void nvme_ns_shutdown(NvmeNamespace *ns)
 
 void nvme_ns_cleanup(NvmeNamespace *ns)
 {
+NvmeZdc *zdc;
+
 if (ns->params.zoned) {
 timer_free(ns->active_timer);
+while ((zdc = QTAILQ_FIRST(&ns->zdc_list))) {
+g_free(zdc);
+}
 g_free(ns->id_ns_zoned);
 g_free(ns->zone_array);
 g_free(ns->zd_extensions);
diff --git a/hw/nvme/nvme.h b/hw/nvme/nvme.h
index 9a54dcdb32..ae65226150 100644
--- a/hw/nvme/nvme.h
+++ b/hw/nvme/nvme.h
@@ -32,6 +32,7 @@ QEMU_BUILD_BUG_ON(NVME_MAX_NAMESPACES > NVME_NSID_BROADCAST - 
1);
 
 typedef struct NvmeCtrl NvmeCtrl;
 typedef struct NvmeNamespace NvmeNamespace;
+typedef struct NvmeZone NvmeZone;
 
 #define TYPE_NVME_BUS "nvme-bus"
 OBJECT_DECLARE_SIMPLE_TYPE(NvmeBus, NVME_BUS)
@@ -90,10 +91,16 @@ static inline NvmeNamespace *nvme_subsys_ns(NvmeSubsystem 
*subs

[PATCH V2 3/4] hw/block/nvme: supply dw1 for aen result

2022-10-21 Thread clay.mayers 
From: Clay Mayers 

cqe.dw1 AEN is sometimes required to convey the NSID of the log page
to read.  This is the case for the zone descriptor changed log
page.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 19 +++
 hw/nvme/trace-events |  2 +-
 include/block/nvme.h |  4 +++-
 3 files changed, 15 insertions(+), 10 deletions(-)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 5b5ceb90c2..4fb85160cc 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1445,6 +1445,7 @@ static void nvme_process_aers(void *opaque)
 result->event_type = event->result.event_type;
 result->event_info = event->result.event_info;
 result->log_page = event->result.log_page;
+req->cqe.dw1 = cpu_to_le32(event->result.nsid);
 g_free(event);
 
 trace_pci_nvme_aer_post_cqe(result->event_type, result->event_info,
@@ -1455,11 +1456,12 @@ static void nvme_process_aers(void *opaque)
 }
 
 static void nvme_enqueue_event(NvmeCtrl *n, uint8_t event_type,
-   uint8_t event_info, uint8_t log_page)
+   uint8_t event_info, uint8_t log_page,
+   uint32_t nsid)
 {
 NvmeAsyncEvent *event;
 
-trace_pci_nvme_enqueue_event(event_type, event_info, log_page);
+trace_pci_nvme_enqueue_event(event_type, event_info, log_page, nsid);
 
 if (n->aer_queued == n->params.aer_max_queued) {
 trace_pci_nvme_enqueue_event_noqueue(n->aer_queued);
@@ -1471,6 +1473,7 @@ static void nvme_enqueue_event(NvmeCtrl *n, uint8_t 
event_type,
 .event_type = event_type,
 .event_info = event_info,
 .log_page   = log_page,
+.nsid   = nsid,
 };
 
 QTAILQ_INSERT_TAIL(&n->aer_queue, event, entry);
@@ -1505,7 +1508,7 @@ static void nvme_smart_event(NvmeCtrl *n, uint8_t event)
 return;
 }
 
-nvme_enqueue_event(n, NVME_AER_TYPE_SMART, aer_info, NVME_LOG_SMART_INFO);
+nvme_enqueue_event(n, NVME_AER_TYPE_SMART, aer_info, NVME_LOG_SMART_INFO, 
0);
 }
 
 static void nvme_clear_events(NvmeCtrl *n, uint8_t event_type)
@@ -5830,7 +5833,7 @@ static uint16_t nvme_ns_attachment(NvmeCtrl *n, 
NvmeRequest *req)
 if (!test_and_set_bit(nsid, ctrl->changed_nsids)) {
 nvme_enqueue_event(ctrl, NVME_AER_TYPE_NOTICE,
NVME_AER_INFO_NOTICE_NS_ATTR_CHANGED,
-   NVME_LOG_CHANGED_NSLIST);
+   NVME_LOG_CHANGED_NSLIST, 0);
 }
 }
 
@@ -6971,7 +6974,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_REGISTER,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -6988,7 +6991,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_VALUE,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -7033,7 +7036,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_REGISTER,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
@@ -7050,7 +7053,7 @@ static void nvme_process_db(NvmeCtrl *n, hwaddr addr, int 
val)
 if (n->outstanding_aers) {
 nvme_enqueue_event(n, NVME_AER_TYPE_ERROR,
NVME_AER_INFO_ERR_INVALID_DB_VALUE,
-   NVME_LOG_ERROR_INFO);
+   NVME_LOG_ERROR_INFO, 0);
 }
 
 return;
diff --git a/hw/nvme/trace-events b/hw/nvme/trace-events
index 337927e607..86c01f8762 100644
--- a/hw/nvme/trace-events
+++ b/hw/nvme/trace-events
@@ -80,7 +80,7 @@ pci_nvme_aer_masked(uint8_t type, uint8_t mask) "type 
0x%"PRIx8" mask 0x%"PRIx8"
 pci_nvme_aer_post_cqe(uint8_t typ, uint8_t info, uint8_t log_page) "type 
0x%"PRIx8" info 0x%"PRIx8" lid 0x%"PRIx8""
 pci_nvme_ns_attachment(uint16_t cid, uint8_t sel) "cid %"PRIu16", 
sel=0x%"PRIx8""
 pci_nvme_ns_attachment_attach(uint16_t cntlid, uint32_t nsid) 
"cntlid=0x%"PRIx16", nsid=0x%"PRIx32""
-pci_nvme_enqueue_event(uint8_t typ, uint8_t info, uint8_t log_page) "type 
0x%"PRIx8" info 0x%"PRIx8" lid 0x%"PRIx8""
+pci_nvme_enqueue_event(uint8_t typ, uint8_t info, uint8_t log_page, uint32_t 
nsid) "type 0x%"PRIx8" info 0x%"PRIx8" lid 0x%"PRIx8" nsi

[PATCH V2 0/4] hw/block/nvme: Implement ZNS finish-zone ZDC AEN

2022-10-21 Thread clay.mayers 
From: Clay Mayers 

ZNS controllers have the option to limit the time a zone can remain in
the active state.  It begins with a background process in the controller
setting the finish-zone-recommended FZR attribute for a zone.  As part of
setting this attribute, the zone's id is added to the namespace's
zone-descriptor-changed (ZDC) log page. If enabled, items added to the
ZDC log page generate a ZDC "asynchronous event notification" AEN. Optionally,
the control can induce a "zone excursion" forcing the zone into the finished
state that also generates a ZDC event.

Zone enabled applications need to properly handle ZDC events. In a real device,
the timeout is many hours making testing an application difficult.
Implemented is the generation of FZR ZDC events to speed up O/S and application
testing.

Added to the zoned NVMe command set is an optional, per-namespace timer
(zoned.finish_time) to set the FZR attr for long-lived active zones; A per
namespace ZDC log page; AEN results to including CQE.DW1 (the NSID of the ZDC
AEN) and generating a ZDC AEN if it's been enabled. Zone excursions are not
modeled.

See section 5.5 of the NVMe Zoned Namespace Command Set Specification v1.1
for more details.

Changes since v1
- Fixed offset length checking in zdc log page
- Moved zdc_event_queued to the patch 4
- Unwatched zdc events in nvme_exit()

Clay Mayers (4):
  hw/block/nvme: add ZONE_FINISH_RECOMMENDED functionality
  hw/block/nvme: add zone descriptor changed log page
  hw/block/nvme: supply dw1 for aen result
  hw/block/nvme: add zone descriptor changed AEN

 docs/system/devices/nvme.rst |   5 +
 hw/nvme/ctrl.c   | 174 +--
 hw/nvme/ns.c |  15 +++
 hw/nvme/nvme.h   |  37 +++-
 hw/nvme/trace-events |   3 +-
 include/block/nvme.h |  14 ++-
 6 files changed, 233 insertions(+), 15 deletions(-)

-- 
2.27.0

[PATCH V2 4/4] hw/block/nvme: add zone descriptor changed AEN

2022-10-21 Thread clay.mayers 
From: Clay Mayers 

If a namespace's param.zoned.finish_time is non-zero,
controllers register with the namespace to be notified
when entries are added to its zone-descriptor-changed
log page.  If the zone-descriptor-changed aen is enabled,
this will cause an AEN to be sent from that controller.

Signed-off-by: Clay Mayers 
---
 hw/nvme/ctrl.c   | 63 +++-
 hw/nvme/ns.c |  1 +
 hw/nvme/nvme.h   | 11 
 include/block/nvme.h |  2 ++
 4 files changed, 76 insertions(+), 1 deletion(-)

diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 4fb85160cc..be6147363d 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1519,6 +1519,52 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_zdc_watch(NvmeCtrl *n, NvmeNamespace *ns, NvmeNotifyFnc fnc)
+{
+NvmeZdcNotify *watcher = g_malloc0(sizeof(*watcher));
+
+watcher->n = n;
+watcher->notify = fnc;
+QTAILQ_INSERT_TAIL(&ns->zdc_watchers, watcher, entry);
+}
+
+static void nvme_zdc_unwatch(NvmeCtrl *n, NvmeNamespace *ns)
+{
+NvmeZdcNotify *watcher;
+
+QTAILQ_FOREACH(watcher, &ns->zdc_watchers, entry) {
+if (watcher->n == n) {
+QTAILQ_REMOVE(&ns->zdc_watchers, watcher, entry);
+break;
+}
+}
+}
+
+static void nvme_zdc_notify(NvmeNamespace *ns)
+{
+NvmeZdcNotify *watcher;
+
+QTAILQ_FOREACH(watcher, &ns->zdc_watchers, entry) {
+(*watcher->notify)(watcher->n, ns);
+}
+}
+
+static void nvme_zdc_aen(NvmeCtrl *n, NvmeNamespace *ns)
+{
+g_assert(n->id_ctrl.oaes & (1 << 27));
+
+if (!NVME_AEC_ZONE_CHANGED(n->features.async_config)) {
+return;
+}
+
+if (!n->zdc_event_queued) {
+n->zdc_event_queued = true;
+nvme_enqueue_event(n, NVME_AER_TYPE_NOTICE,
+NVME_AER_INFO_NOTICE_ZONE_DESC_CHANGED,
+NVME_LOG_CHANGED_ZONE, ns->params.nsid);
+}
+}
+
 static void nvme_zdc_list(NvmeNamespace *ns, NvmeZoneIdList *zlist, bool reset)
 {
 NvmeZdc *zdc;
@@ -1554,6 +1600,7 @@ static void nvme_check_finish(NvmeNamespace *ns, 
NvmeZoneListHead *list)
 zdc->zone = zone;
 zone->zdc_entry = zdc;
 QTAILQ_INSERT_TAIL(&ns->zdc_list, zdc, entry);
+nvme_zdc_notify(ns);
 }
 } else if (zone->finish_ms != INT64_MAX) {
 timer_mod_anticipate(ns->active_timer, zone->finish_ms);
@@ -4727,6 +4774,14 @@ static uint16_t nvme_changed_zones(NvmeCtrl *n, uint8_t 
rae, uint32_t buf_len,
 return NVME_INVALID_NSID | NVME_DNR;
 }
 nvme_zdc_list(ns, &zlist, !rae);
+if (!rae) {
+n->zdc_event_queued = false;
+nvme_clear_events(n, NVME_AER_TYPE_NOTICE);
+/* send a new aen if there are still zdc entries */
+if (!QTAILQ_EMPTY(&ns->zdc_list)) {
+nvme_zdc_notify(ns);
+}
+}
 
 trans_len = MIN(sizeof(zlist) - off, buf_len);
 
@@ -5815,6 +5870,7 @@ static uint16_t nvme_ns_attachment(NvmeCtrl *n, 
NvmeRequest *req)
 return NVME_NS_NOT_ATTACHED | NVME_DNR;
 }
 
+nvme_zdc_unwatch(n, ns);
 ctrl->namespaces[nsid] = NULL;
 ns->attached--;
 
@@ -7542,7 +7598,7 @@ static void nvme_init_ctrl(NvmeCtrl *n, PCIDevice 
*pci_dev)
 
 id->cntlid = cpu_to_le16(n->cntlid);
 
-id->oaes = cpu_to_le32(NVME_OAES_NS_ATTR);
+id->oaes = cpu_to_le32(NVME_OAES_NS_ATTR | NVME_OAES_ZDC);
 id->ctratt |= cpu_to_le32(NVME_CTRATT_ELBAS);
 
 id->rab = 6;
@@ -7659,6 +7715,10 @@ void nvme_attach_ns(NvmeCtrl *n, NvmeNamespace *ns)
 
 n->dmrsl = MIN_NON_ZERO(n->dmrsl,
 BDRV_REQUEST_MAX_BYTES / nvme_l2b(ns, 1));
+
+if (ns->params.fto) {
+nvme_zdc_watch(n, ns, nvme_zdc_aen);
+}
 }
 
 static void nvme_realize(PCIDevice *pci_dev, Error **errp)
@@ -7721,6 +7781,7 @@ static void nvme_exit(PCIDevice *pci_dev)
 for (i = 1; i <= NVME_MAX_NAMESPACES; i++) {
 ns = nvme_ns(n, i);
 if (ns) {
+nvme_zdc_unwatch(n, ns);
 ns->attached--;
 }
 }
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index 25cd490c99..5629b61302 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -241,6 +241,7 @@ static void nvme_ns_zoned_init_state(NvmeNamespace *ns)
 QTAILQ_INIT(&ns->closed_zones);
 QTAILQ_INIT(&ns->full_zones);
 QTAILQ_INIT(&ns->zdc_list);
+QTAILQ_INIT(&ns->zdc_watchers);
 
 zone = ns->zone_array;
 for (i = 0; i < ns->num_zones; i++, zone++) {
diff --git a/hw/nvme/nvme.h b/hw/nvme/nvme.h
index ae65226150..5499105e7b 100644
--- a/hw/nvme/nvme.h
+++ b/hw/nvme/nvme.h
@@ -91,6 +91,14 @@ static inline NvmeNamespace *nvme_subsys_ns(NvmeSubsystem 
*subsys,
 #define NVME_NS(obj) \
 OBJECT_CHECK(NvmeNamespace, (obj), TYPE_NVME_NS)
 
+typedef void (*NvmeNotifyFnc)(NvmeCtrl *n, NvmeNamespace *ns);
+
+typedef struct Nvme

[PATCH V2 1/4] hw/block/nvme: add ZONE_FINISH_RECOMMENDED functionality

2022-10-21 Thread clay.mayers 
From: Clay Mayers 

Adds ns.param.zoned.finish_time, which sets the number of
seconds a zone can remain active before the zone attribute
ZONE_FINISH_RECOMMENDED is set.

This requires scanning the exp open, imp open and closed lists
of zones whenever a zone is marked as requiring finishing.  The
expectation is these lists will be short (10s of items) allowing a
simpler implementation than keeping the lists sorted.  It also
keeps the overhead during the exception of a timeout instead of
when zones change state between open and closed. For use cases
where this isn't true, finish_time should be 0 to disable this
feature (the default).

Signed-off-by: Clay Mayers 
---
 docs/system/devices/nvme.rst |  5 +
 hw/nvme/ctrl.c   | 35 +++
 hw/nvme/ns.c |  8 
 hw/nvme/nvme.h   | 18 ++
 4 files changed, 62 insertions(+), 4 deletions(-)

diff --git a/docs/system/devices/nvme.rst b/docs/system/devices/nvme.rst
index 30f841ef62..1cb0ef844c 100644
--- a/docs/system/devices/nvme.rst
+++ b/docs/system/devices/nvme.rst
@@ -212,6 +212,11 @@ The namespace may be configured with additional parameters
   the minimum memory page size (CAP.MPSMIN). The default value (``0``)
   has this property inherit the ``mdts`` value.
 
+``zoned.finish_time=UINT32`` (default: ``0``)
+  Set the time in seconds for how long a zone can be active before setting the
+  zone attribute ``Zone Finish Recommended``.  The default value (``0``)
+  disables this feature.
+
 Metadata
 
 
diff --git a/hw/nvme/ctrl.c b/hw/nvme/ctrl.c
index 87aeba0564..d7e9fae0b0 100644
--- a/hw/nvme/ctrl.c
+++ b/hw/nvme/ctrl.c
@@ -1516,6 +1516,38 @@ static void nvme_clear_events(NvmeCtrl *n, uint8_t 
event_type)
 }
 }
 
+static void nvme_check_finish(NvmeNamespace *ns, NvmeZoneListHead *list)
+{
+int64_t now = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL);
+NvmeZone *zone;
+
+QTAILQ_FOREACH(zone, list, entry) {
+if (zone->finish_ms <= now) {
+zone->finish_ms = INT64_MAX;
+zone->d.za |= NVME_ZA_FINISH_RECOMMENDED;
+} else if (zone->finish_ms != INT64_MAX) {
+timer_mod_anticipate(ns->active_timer, zone->finish_ms);
+}
+}
+}
+
+void nvme_finish_needed(void *opaque)
+{
+NvmeNamespace *ns = opaque;
+
+nvme_check_finish(ns, &ns->exp_open_zones);
+nvme_check_finish(ns, &ns->imp_open_zones);
+nvme_check_finish(ns, &ns->closed_zones);
+}
+
+void nvme_set_active_timeout(NvmeNamespace *ns, NvmeZone *zone)
+{
+if (ns->fto_ms) {
+zone->finish_ms = qemu_clock_get_ms(QEMU_CLOCK_VIRTUAL) + ns->fto_ms;
+timer_mod_anticipate(ns->active_timer, zone->finish_ms);
+}
+}
+
 static inline uint16_t nvme_check_mdts(NvmeCtrl *n, size_t len)
 {
 uint8_t mdts = n->params.mdts;
@@ -1791,6 +1823,7 @@ static uint16_t nvme_zrm_finish(NvmeNamespace *ns, 
NvmeZone *zone)
 
 /* fallthrough */
 case NVME_ZONE_STATE_EMPTY:
+zone->d.za &= ~NVME_ZA_FINISH_RECOMMENDED;
 nvme_assign_zone_state(ns, zone, NVME_ZONE_STATE_FULL);
 return NVME_SUCCESS;
 
@@ -1891,6 +1924,7 @@ static uint16_t nvme_zrm_open_flags(NvmeCtrl *n, 
NvmeNamespace *ns,
 
 if (act) {
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 }
 
 nvme_aor_inc_open(ns);
@@ -3619,6 +3653,7 @@ static uint16_t nvme_set_zd_ext(NvmeNamespace *ns, 
NvmeZone *zone)
 return status;
 }
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 zone->d.za |= NVME_ZA_ZD_EXT_VALID;
 nvme_assign_zone_state(ns, zone, NVME_ZONE_STATE_CLOSED);
 return NVME_SUCCESS;
diff --git a/hw/nvme/ns.c b/hw/nvme/ns.c
index 62a1f97be0..b577f2d8e0 100644
--- a/hw/nvme/ns.c
+++ b/hw/nvme/ns.c
@@ -322,6 +322,11 @@ static void nvme_ns_init_zoned(NvmeNamespace *ns)
 ns->id_ns.nsfeat &= ~0x4;
 }
 
+ns->fto_ms = ns->params.fto * INT64_C(1000);
+if (ns->fto_ms) {
+ns->active_timer = timer_new_ms(QEMU_CLOCK_VIRTUAL, nvme_finish_needed,
+ns);
+}
 ns->id_ns_zoned = id_ns_z;
 }
 
@@ -338,6 +343,7 @@ static void nvme_clear_zone(NvmeNamespace *ns, NvmeZone 
*zone)
 nvme_set_zone_state(zone, NVME_ZONE_STATE_CLOSED);
 }
 nvme_aor_inc_active(ns);
+nvme_set_active_timeout(ns, zone);
 QTAILQ_INSERT_HEAD(&ns->closed_zones, zone, entry);
 } else {
 trace_pci_nvme_clear_ns_reset(state, zone->d.zslba);
@@ -521,6 +527,7 @@ void nvme_ns_shutdown(NvmeNamespace *ns)
 void nvme_ns_cleanup(NvmeNamespace *ns)
 {
 if (ns->params.zoned) {
+timer_free(ns->active_timer);
 g_free(ns->id_ns_zoned);
 g_free(ns->zone_array);
 g_free(ns->zd_extensions);
@@ -644,6 +651,7 @@ static Property nvme_ns_props[] = {
 DEFINE_PROP_SIZE("zoned.zrwafg", NvmeNamespace, params.zrwafg, -1),